An Update On LBMA Refinery Statistics & GFMS

Last year I published two posts featuring the 2013 refinery production statistics of the London Bullion Market Association (LBMA) Good Delivery Refiners (gold and silver). The first post, “Moving the goalposts….The LBMA’s shifting stance on gold refinery production statistics“, looked primarily at how, in May 2015, the LBMA had first published a 2013 gold refining output total for its member refineries of 6601 tonnes, only to dramatically reduce this total to 4,600 tonnes in early August 2015, a few days after the 6,601 tonnes figure was mentioned on this blog.

Not only that, but the early August LBMA revisions also introduced a dubious comparison approach of comparing the LBMA refiners’ output to the rigid and incomplete GFMS data model of total global gold supply (mining output, scrap recycling and net hedging). There were also unexplained changes by the LBMA to the silver output figures at that time in early August 2015.

The assertion at the time, concluding the article was:

“Could it be that this 6,601 tonne figure included refinery throughput for the huge number of London Good Delivery gold bars extracted from gold ETFs and LBMA and Bank of England vaults and converted into smaller gold bars in 2013, mainly using LBMA Good Delivery Swiss gold refineries? And that maybe this 6,601 tonne figure stood out as a statistical outlier for 2013 which no one wanted to talk about?"

I heard some feedback from journalistic sources that the LBMA did not want to discuss this 6,601 tonnes figure when asked, so make of that what you will.

My second post on the topic,"From Good Delivery bars to Kilobars – The Swiss Refineries, the GFMS data, and the LBMA“, published in November 2015, investigated the trends behind this 6,601 tonne number and looked at the phenomenon of Good Delivery bars being processed through the Swiss gold refineries in 2013, the gold withdrawals from the London-based gold ETFs and the Bank of England, the massive shipments of gold from the UK to Switzerland in 2013, and how GFMS and the World Gold Council attempted to explain (or not explain) the conversion of large bars to smaller kilo bars given that their supply-demand models don’t capture this activity.

The LBMA has now released 2014 refinery production of its member refineries for gold and silver (in a press release and a brochure) hence the reason for revisiting this topic. Yet again, there are inconsistencies and a number of shocking errors in the details that the LBMA has released, but also a few insights into 400 oz bar to kilobar conversions.

Schoolboy Errors

On 18 May, the LBMA issued a press release titled “4,921 tonnes of gold production in 2014 – LBMA GD refiners" covering gold and silver refinery production output of its refining members.

For gold, the press release reads:

“In 2014, total refined gold production by the refiners on the LBMA’s Gold Good Delivery List was 4,921 tonnes. This is above world mine production of 4,394 tonnes (source: Thomson Reuters GFMS) with the difference reflecting the recycling of material by LBMA GD refiners converting large 400 oz bars into kilobars.“

Note that this is the first time, as far as I know, that the LBMA has even broached the subject of conversion of large 400oz bars (held by central banks and ETFs) into kilobars.

However, the first major error in this LBMA press release is its claim that “world mine production" was 4,394 tonnes in 2014, the source of which it attributes to GFMS. This is not true. The 4,394 tonnage figure from the “GFMS World Gold Survey 2016" (published 31 March 2016), refers to the ‘total’ global gold supply in 2014.

Total Supply = Mine Production  +  Scrap  +  Net Hedging Supply

GFMS 2014 data: Mine Production 3131  +  Scrap 1158  + Net Hedging 104   =   4394

See supply and demand table on page 7 in the GFMS report’s “Summary and Price Outlook" section.

The LBMA has therefore taken a figure comprised of Mine Production + Scrap + Net Hedging Supply and proceeded to call it ‘World Mine Production". Although it’s obviously a schoolboy error, it shows that whoever wrote the press release, proof-read and published the press release (and brochure) wasn’t alert to these definitions.

The second flaw in the LBMA statement is that it continues to compare real production output of its member refiners to a ‘model’ published by GFMS which includes estimates for items like net hedging, and that, most importantly, totally ignores 400oz bar to kilobar bar conversions. This is like comparing apples to oranges.

When a consultancy such as GFMS first publishes full year data covering 2014, it does so in early 2015. Then when 2016 rolls around, the 2016 report, which primarily covers 2015 data, will also include prior year data such as 2014 but often there are small tweaks and changes in some of the data components, presumably as data in the model gets updated. This is understandable.  Going back to the GFMS ‘Full year report for 2015’ (published in March 2015), there are slight differences in the 2014 figures published in 2015 compared to the  March 2016 version, mostly because of a slightly lower scrap supply figure.

2014 data (in 2015) :  Mine Production 3133  +  Scrap 1125  + Net Hedging 103   =   4362

However, page 10 of the GFMS 2015 gold report, makes it abundantly clear that there are 3 components to total global gold supply in its model:

“Mine production increased for a sixth successive year in 2014, rising by 2% to a record volume of 3,133 tonnes.”

“producer hedging generated 103 tonnes of accelerated supply in 2014” 

“Global scrap supply retreated 13% in 2014 to a seven-year low of 1,125 tonnes, as a result of a weaker dollar gold price and an improved economic environment”

LBMA GD refiners converting large 400 oz bars into kilobars

The most interesting part of the LBMA May 2016 press release is that it says the difference between its 4,921 tonnes refiner output and the GFMS total supply figure of 4,394 tonnes reflects “the recycling of material by LBMA GD refiners converting large 400 oz bars into kilobars". Therefore, the LBMA is stating that 527 tonnes of good delivery bars were recast into kilobars during 2014. The majority of these conversions would be at the large Swiss gold refineries, which are the largest gold refineries in the world and are all members of the LBMA Good Delivery List. In fact, 3 of the 4 big Swiss gold refineries are 3 of the 5 members of the LBMA refinery referee panel.

Note, GFMS discuss their scrap gold figure in terms of components such as e-waste recycling, jewellery scrap, dental, coins, electronic scrap etc, so again, these sources of scrap have nothing to do with 400oz bars.

Notwithstanding the fact that comparing LBMA refiners’ actually output to a GFMS figure where the component for net hedging is estimated and represents ‘paper’ delta hedging, this 400oz bar to Kg bar conversion is not part of the GFMS figures. The LBMA’s comparison to a GFMS total supply figure that includes delta hedging doesn’t seem to bother to analyse the hedging methodology to see if it represents physical gold coming to the market during that specific year (2014). For example, the 2 largest components of the GFMS’ net hedging increase in 2014 are attributed to Polyus Gold in Russia (using derivatives such as forward sales and collars) and Fresnillo. GFMS even have another term for this net hedging and they call it “changes in lending" in a pie-chart diagram on page 52 of their 2015 full year report.

Interestingly, in its pie chart on page 52, GFMS labels the total supply figure of 4362 tonnes (pre-vised figure) as “Transformed / Transferred“. So again, the GFMS model does not take account of 400oz bars to Kg bar conversions, which the dogs on the street could tell you is a form of ‘transformation’.

Page 53 of the GFMS 2015 report introduces a supply source of ETF drawdowns, which it estimates to be 160 tonnes. GFMS then adds this ETF drawdown to its 4362 tonnes figure to arrive at 4,521 tonnes of visible supply of gold to the market during 2014, which is still (exactly) 400 tonnes different to the 4,921 figure from the LBMA’s refiners. So there is a 400 tonne delta that GFMS data doesn’t explain.

Excluding ‘net hedging’, which has a tenuous connection to physical supply changes, the total quantity of unexplained gold refinery output of the LBMA member refineries in 2014 was 663 tonnes.

i.e.  4921  –  3133  –  1125 =  663 tonnes

If 160 tonnes of this was supplied by ETF drawdowns, then there is still ~ 527 tonnes of unexplained 400 oz bar to kilobar conversions for 2014. The only possible source of such as supply is from central bank vaults, i.e. unreported central bank gold sales, or more likely leased central bank gold that has gone to Switzerland and then on to Asia. Not all of the 160 tonnes of ETF drawdowns during 2014 would necessarily have been sent to refineries. Some could have stayed in vaults in London and elsewhere, or else been used to repay gold borrowings from central banks. Therefore, the 527 tonnes of large to small bar conversion refining by LBMA members could be even higher still.

LBMA Silver statement: More than vs. Less than

Turning to silver, the LBMA press release is even stranger and states:

“For silver, refined production by refiners on the LBMA’s Silver Good Delivery list in 2014 was estimated to be 31,131 tonnes, above the 33,025 tonnes of mine production in 2014 (source: Thomson Reuters GFMS)"

This LBMA statement is completely illogical. How can the number 31,131 be above the number 33,025? Only in the reverse world of the LBMA. Notice again the comparison of refinery output statistics to a GFMS supply model.

The LBMA states erroneously in its press release that ‘mine production in 2014‘  was 33,205 tonnes, equivalent to approximately 1,067.5 million ounces. The “GFMS Full year Silver report for 2016" provides silver supply in millions of ounces and says that “Mine Supply for 2014" was 868.3 ounces, which is 27,0007.7 tonnes, and that “Total silver supply for 2014" was 1053.3 million ozs, which is 32,762 tonnes. The above wording also uses the word ‘estimated to be 31,131". this is mis-leading because there is no need for estimates of LBMA refinery production because every refinery on the Good Delivery ‘Lists’ provides the exact real production figures to the LBMA.

Therefore, for silver, the LBMA press release is wrong on 5 counts, a) it claims one number is larger than another when it’s not, b) it claims mine supply was far higher than its was, c) its figure is not even the same as in the GFMS report for ‘total supply’, and d) it compares refining output to a GFMS model, which is like comparing apples to oranges, and e) there is no need for estimates of refinery production when every refinery on the Good Delivery ‘Lists’ provides the exact real production figures to the LBMA.

The Updated LBMA Brochure

Two weeks prior to issuing the above press release, the LBMA updated its standard brochure, which is titled " A guide to The London Bullion Market Association May 2016". The May 2016 version can be seen here. This pdf brochure file was created on 5 May 2016, and modified on 10 May 2016. The brochure also contains a section about refinery production statistics on Page 3, where it states:

“Total refined gold production by the refiners on the List was estimated to be 4,921 tonnes in 2014, owing to recycling of scrap material, above world mine production of 4,394 tonnes (source: Thomson Reuters GFMS).

For silver, refined production by listed refiners in 2014 was estimated to be 31,131 tonnes, above the 33,025 tonnes of mine production in 2014. (source: Thomson Reuters GFMS)."

Notice that the Silver sentence is the same as in the press release (and wrong). The Gold sentence differs to that in the press release.

Lets compare the LBMA press release statement to the LBMA brochure statement:

Press release:

“In 2014 total refined gold production by the refiners on the LBMA’s Gold Good Delivery List was 4,921 tonnes.This is above world mine production of 4,394 tonnes (source: Thomson Reuters GFMS) with the difference reflecting the recycling of material by LBMA GD refiners converting large 400 oz bars into kilobars."

Brochure:

“Total refined gold production by the refiners on the List was estimated to be 4,921 tonnes in 2014, owing to recycling of scrap material, above world mine production of 4,394 tonnes (source: Thomson Reuters GFMS).

In the brochure, the world mine production figure schoolboy error is still present, which should state total GFMS global gold supply. More fundamentally, there is no reference to the conversion of 400 oz bars into kilobars in this brochure, and the difference between the LBMA and GFMS figures is attributed to ‘recycling of scrap material". However, this explanation is a mathematical impossibility because scrap recycling of 1158 tonnes is already in the 4,394 tonnes figure.  This is a real clanger of an error, and completely negates any trust in the statement.

For gold therefore, the LBMA brochure is wrong on 4 counts, a) it claims mine supply was far higher than its was, b) it claims a difference is due to a supply component that’s already part of a figure it provides (i.e. the 4,394 tonnes figure), c) it uses a misleading ‘estimated to be phrase’, but as explained, there is no need for estimates of refinery production when every refinery on the Good Delivery ‘Lists’ provides the exact real production figures to the LBMA. and d) it compares refining output to a rigid and incomplete GFMS model, which is like comparing apples to oranges.

Conclusion

The critical omission in the brochure is the lack of reference to the delta being due to “converting large 400 oz bars into kilobars".  This is between 527 and 665 tonnes of gold we’re talking about here, more than the official gold reserves of all but the largest central bank and ETF holders.

Weirdly, since the brochure was created on 5 May and last modified on 10 May, it was written and finalised before the press release went out on 18 May. So why would the LBMA write one misleading explanation for the difference in the brochure that deflects attention away from the conversion of central bank and ETF bars into kilobars, and then seem to be slightly more truthful in the press release 2 weeks later when mentioning conversion of 400oz bars into kilobars? Perhaps the press release wasn’t read by the Ministry of Truth before being released.

Overall, this series of communications from the LBMA leaves a lot to be desired, and undermines trust in LBMA issued data. It will be interesting to see whether the LBMA will surreptitiously alter its press release and brochure, or issue a correction and clarification, or leave the errors as is in the respective documents.